Key findings
- This note estimates the neutral rate for New Zealand using a Vector Error Correction Model (VECM). The neutral interest rate is the rate of interest at which monetary policy is neither expansionary or contractionary. The neutral rate is not directly observable and hence must be estimated. The VECM relies on the long-run relationship between short- and long-term ex-ante real interest rates to explain changes in the neutral rate.
- The estimated real neutral rate shows a downward trend following 2008 starting from approximately 4 percent, reaching negative values during the onset of the pandemic in 2020, and since then, steadily rising to positive values at the end of the estimation period in 2023. Long-term estimates of the nominal neutral rate under this framework suggests that monetary policy was contractionary from 2022Q2 to 2023Q4.
- The VECM estimation allows for the decomposition of neutral rate changes, which is a valuable tool to understand the drivers of changes in the neutral rate. Interest rates dynamics explain most of the change in neutral rates between 1997 and 2023. However, I also find the following factors contribute to the decline in neutral rates: lower productivity, appreciating nominal exchange rates, and to a lesser degree, a decrease in real investment and higher public debt.
Why we did this research
This note estimates New Zealand’s neutral rate using a novel-for-New Zealand multivariate regression approach not currently included in the Reserve Bank’s range of neutral rate models. Specifically, the Vector Error Correction Model (VECM) framework introduced in this Note provides an empirical estimate for the neutral rate by employing a multivariate Beveridge-Nelson decomposition. The estimated neutral rate under this framework can be decomposed to assess the contribution of each data series to the change in neutral rate over the estimation period. This approach better captures interdependencies between the economic variables compared to structural models, allowing for data-driven relationships and considering a larger number of variables to decompose the movements in the neutral rate. The VECM r* is broadly consistent with existing RBNZ neutral model estimates, demonstrating this methodology's value as a complementary tool for policymaking and analysis.
What data have we used?
| Variables | Frequency | Transformation | Source |
|---|---|---|---|
| NZ 3-month bill rate | Daily | Quarterly average, level/ first difference | Reuters |
| NZ 10 years swap rate | Daily | Quarterly average, level/ first difference | Reuters |
| Public debt as percentage of GDP | Annual | Interpolated, first difference | International Monetary Fund |
| Real private investment | Quarterly | First log difference | Stats NZ |
| Total hours worked | Quarterly | First log difference | Stats NZ |
| Economic policy uncertainty | Monthly | Quarterly average, first difference | Ali, S., Badshah, I., Demirer, R., Hegde, P. (2022) |
| Current account as percentage of GDP | Quarterly | First difference | Stats NZ |
| Number of employees (thousands) | Quarterly | First log difference | Stats NZ |
| Private consumption | Quarterly | First log difference | Stats NZ |
| NZSIM Total factor productivity | Quarterly | First difference | RBNZ |
| Trade weighted index nominal exchange rate | Quarterly | First log difference | RBNZ |
| Unemployment rate | Quarterly | First difference | Stats NZ |
| Consumer price index | Quarterly | First difference | Stats NZ |